Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Deep learning job postings have collapsed in the past six months

twitter.com

Post reply on HN
Print view View original

211–220 of 274 posts

Re: Deep learning job postings have collapsed in the past six months

#211

Post by ghaff »

alexslobodnik wrote: post #187
barkingcat wrote: post #168

Earlier quoted context omitted.

This can be applied as "nobody listens to the people who actually do the work" as in company hires ML/AI experts to analyze purchase records and service records, and spits back out trends that the service front line workers (tier 1) already knew dead solid. Then the company doesn't listen to either group of people (neither tier 1 sales/support people, nor the ML people) and then fires / shuts down the entire division…

Or it could be that a lot of data is wrong. It may be "technically" correct, ie the table in a database produces X. It is no surprise that executives would ignore what the "data" says because they don't trust it. A lot of time they are right to ignore it. I've seen tables say X, but there was some flaw up the capture stack. Very few data analyst have the broad based knowledge and dedication needed to trace the data s…

Probably the biggest part of a lot of projects is cleaning up and normalizing the data (just like it was with data warehousing last century). A lot of high power data scientists don't want to hear that--especially if they're research-y.

Re: Deep learning job postings have collapsed in the past six months

#212

Post by at-fates-hands »

eric_b wrote: post #28

I've worked in lots of big corps as a consultant. Every one raced to harness the power of "big data" ~7 years ago. They couldn't hire or spend money fast enough. And for their investment they (mostly) got nothing. The few that managed to bludgeon their map/reduce clusters in to submission and get actionable insights discovered... they paid more to get those insights than they were worth! I think this same thing is ha…

The company I work for is a large health care company. I started in robotic automation about a year ago. The company said its next three huge initiatives would be:

1) AI 2) Machine Learning 3) Robotic Process Automation

They felt RPA would help them stay more competitive since there are tons of smaller health care companies who are moving faster and innovating faster because they're not buying up companies and having to integrate all their technology at a sloth's pace. They thought RPA would be a way to mitigate these issues.

18 months later and the one manager, director and VP in my org has all but said they don't care about RPA, all their money is going into ML and AI. Even though in all the presentations I've seen them put on, its all blue skies and BS about "IF we had this, we COULD do this." Nothing concrete at all about how the plan to use ML to increase profit margins or reduce overhead.

Right now, our team is basically an afterthought in the company and I'm already starting to interview elsewhere with the knowledge at some point, they're going to kill my team and cut everybody loose.

Re: Deep learning job postings have collapsed in the past six months

#214

Post by apohn »

xmprt wrote: post #174
apohn wrote: post #84

Earlier quoted context omitted.

>If you think about it, that's the natural outcome. Why? Because people in corporations don't have the incentive to benefit the business but to progress their careers and that's done through meeting the goals for their position and make their upper ups progress with their careers too. This is one of the reasons I roll my eyes whenever I read something like "McKinsey says 75% of Big Data/AI/Buzzword projects do not de…

Why do you roll your eyes? Isn't it a useful metric to know that most of the projects that are hiring these buzzword technologies are destined to fail (whether that's because the problem space wasn't fit for ML or whether management went on a hiring spree to pump their resume)?

I dislike these types headlines because I've interacted with a lot of people who see this as evidence that ML/AI is BS and destined to fail. The reality is that a project with unrealistic expectations is going to fail regardless of it being an AI project or somebody baking a loaf of bread at home.

It's important to understand why stuff fails. That's the only way to stop things from failing in the future and make sure people are on the right path to not failing. If a large number project failures are management failures, it's useful to know that. Otherwise you try to fix everything except the management structure.

Re: Deep learning job postings have collapsed in the past six months

#215

Post by The_rationalist »

sooheon wrote: post #34
lacker wrote: post #29

Earlier quoted context omitted.

Could you give an example of a major task that you think the state of the art could be trivially improved on with the xlnet approach?

Long (>2048 tokens) sequences. But GP is too focused on hyping XLNet for some reason. There are much more elegant attempts at improving the transformer architecture in just the past 8 months: Reformer, Performer, Macaron Net, and my current pet paper, Normalized Attention Pooling ( https://arxiv.org/abs/2005.09561 ).

But GP is too focused on hyping XLNet for some reason. Yeah some reasons, it might be because of stars alignment and could variate inversely with the weather. Or it might be because it's the paper with the biggest number of first place at SOTA leaderboards? https://paperswithcode.com/paper/xlnet-generalized-autoregre...

I've queried most of your examples on the SOTA database that is paperswithcode.com and they have almost zero results. You illustrate the problem, if researchers like you don't even know the general SOTA, how can it be expected to be beaten? But beyond scientists ignorance there is also the problem of models not submitting their results to paperswithcode.com or not testing them extensively but only on niche benchmarks. This second behavior sentence such potentially promising models to remain unknown and therefore mostly irrelevant.

It's always remarkable how one can be a smart researcher and yet not adjust its behavior to be rational regarding those two flaws (not seeking SOTA knowledge, and not promoting SOTA knowledge; READ, WRITE)

Re: Deep learning job postings have collapsed in the past six months

#217

Post by morvita »

mumblemumble wrote: post #99
baron_harkonnen wrote: post #51

Earlier quoted context omitted.

> they paid more to get those insights than they were worth! This understates how awful ML is at many of these companies. I've seen quite a few companies that rushed to hire teams of people with a PhD in anything that barely made it through a DS/ML boot camp. To prove that they're super smart ML researchers without fail these hires rush to deploy a 3+ layer MLP to solve a problem that need at most a simple regression…

My sense is that the original sin here is conflating data science with machine learning. A good data scientist might choose to use machine learning to accomplish their job. Or they might find that classical statistical inference is the better tool for the task at hand. A good data scientist, having built this model, might choose to put it into production. Or they might find that a simple if-statement could do the job…

This matches my experience exactly.

About 5 years ago I was coming out of academia with a PhD in chemistry wanting to get into tech. Someone pointed me toward data science and I was immediately pulled in by the potential for deep insights and working with interesting data sets. After getting into an industry research position, I was quickly disillusioned by the talk from our leadership of how we needed to become an "AI company" and discovered that was I really wanted to be doing was classic algorithm development, not data science.

Now, I've see so much poorly implemented, unnecessary machine learning applied to problems that didn't need it that I first assume that any machine learning project is a bad technical decision until proven otherwise. I've happily moved into an engineering role building interesting pipelines.

Re: Deep learning job postings have collapsed in the past six months

#218

Post by eaton »

apohn wrote: post #214
xmprt wrote: post #174

Earlier quoted context omitted.

Why do you roll your eyes? Isn't it a useful metric to know that most of the projects that are hiring these buzzword technologies are destined to fail (whether that's because the problem space wasn't fit for ML or whether management went on a hiring spree to pump their resume)?

I dislike these types headlines because I've interacted with a lot of people who see this as evidence that ML/AI is BS and destined to fail. The reality is that a project with unrealistic expectations is going to fail regardless of it being an AI project or somebody baking a loaf of bread at home. It's important to understand why stuff fails. That's the only way to stop things from failing in the future and make sure…

Statistically, though, that's what the people in this thread are saying — that the majority of the projects in ML/AI are destined to fail because they're BS with unrealistic goals.

"Personalization" is in a similar place for digital publishing; everyone wants it, products and services carry big price tags, and few organizations want to invest in foundational work or simple, iterative improvements. So they swing for the stars with unrealistic goals like "micro-targeted messaging perfectly tailored to every visitor, no matter where they are in the customer journey" and the results are predictable…

I take the increasingly grim accounts of project failure rates from analyst firms as a good sign — they can be used to sober up executives with unrealistic dreams.

Re: Deep learning job postings have collapsed in the past six months

#219

Post by actusual »

softwaredoug wrote: post #197

There's a lot of what I call "model fetishism" in machine learning. Instead of focusing our energies on the infrastructure and quality of data around machine learning, there's eagerness to take bad data to very high-end models. I've seen it again and again at different companies, usually always with disastrous consequences. A lot of these companies would do better to invest in engineering and domain expertise around…

This is why my interview question focuses around applying linear regression to a complex domain. It weeds out an enormous number of candidates.

There are 5 ML models that we maintain where I work, and none of then are more complicated than linear regression or random forests. Convincing me to use something more complex would take an enormous amount of evidence. Domain knowledge is king.

Re: Deep learning job postings have collapsed in the past six months

#220

Post by wokwokwok »

mumblemumble wrote: post #83
austinl wrote: post #74

Earlier quoted context omitted.

I've heard this happen in a lot of places — companies want to be "data-driven", but then leadership simply ignores the data. I think being data-driven is something that is built into company culture, or otherwise it's too easy to just ignore the results and ship. The place I currently work is data-driven (perhaps to a fault). Every change is wrapped behind an experiment and analyzed. Engineers play a major role in th…

Imagine what it must be like for the senior leadership of an established company to actually become data-driven. All of a sudden the leadership is going to consent to having all of their strategic and tactical decision-making be questioned by a bunch of relatively new hires from way down the org chart, whose entire basis for questioning all that expertise and business acumen is that they know how to fiddle around wit…

That just reads as leaders who are high on cognitive bias.

...probably true to some extent, but not all leaders are self important ass hats who refuse to acknowledge they are simply “making decisions” not “making good decisions”. Most leaders are doing the best they can (often even very well) with the insights available to them.

I don’t think most data teams are really at fault; they’re just doing what they’re told.

The problem imo lies with the analysts who fail to do anything useful with the data they’re given, and demand constant changes from the data team because they want to deliver silver bullet results to the leadership level.

That’s the problem layer; people who want to be important but have nothing to offer, whipping their data team to produce rubbish and then blaming them for either a) not producing anything fast enough or b) not making the numbers big enough.

Post reply on HN